Forecasting and analysis of the effect of lifestyle on cognitive dysfunction induced by occupational aluminum exposure based on Bayesian networks

Abstract

Objectives: To evaluate the risk of cognitive impairment in workers with plasma aluminum concentrations and lifestyles using a Bayesian network (BN).

Methods: In 2019, 476 male workers in the Shanxi Aluminum factory were investigated. We measured plasma aluminum concentrations in workers by inductive coupled plasma mass spectrometry (ICPMS) and tested workers' cognitive function by the MoCA scale. We collected the data of lifestyle by the occupational Workers' Health questionnaire and express the influence of lifestyle on cognition by the OR value (95 %CI) of logistic regression. A Bayesian network model was used to predict the risk of cognitive dysfunction.

Results: The subjects were divided into a cognitively normal group and cognitively impaired group according to MoCA scores. There were statistically significant differences in age, education level, alcohol consumption, physical exercise, reading, aluminum length of service and blood aluminum concentration between the two groups (P < 0.05). The plasma aluminum concentration in the cognitive impairment group was 1.68 times higher than that in the cognitive normal group. Four groups were established according to the quartile of blood aluminum concentration of the subjects, namely, Group Q1 (<14.95 μg/L), Q2 group (14.95-32.96 μg/L), Q3 group (32.96-56.62 μg/L), and Q4 group (>56.62 μg/L). Binary logistic regression analysis showed that in the adjustment variable Model2, drinking, short sleep, long sleep, and mobile phone use increased the risk of cognitive impairment by 1.505(0.99,2.289), 1.269(0.702,2.295), 1.125(0.711,1.781) and 1.19(0.779,1.82), respectively, compared with their reference values. The risk of cognitive impairment from reading and exercise was 0.7(0.398,1.232) and 0.787(0.51,1.217), respectively, compared with those of no reading and no exercise. The risk of cognitive impairment of blood aluminum concentration in the Q2, Q3, and Q4 groups was 2.103(1.092,4.051), 1.866(0.955,3.644) and 3.679(1.928,7.020), respectively, compared with that in the Q1 group. Compared with age <40 , the risk of cognitive impairment of age ≥40 was 2.515(1.508,4.193) (P < 0.05). Bayesian network model results showed that if all participants had plasma aluminum concentrations higher than Q4, the prevalence of cognitive impairment was 54.5 %. The prevalence of cognitive impairment was 75.0 % if all participants had plasma aluminum levels above Q4, were older than 40, smoked, drank alcohol, used a cell phone for more than 2 h, slept for more than 8 h, did not exercise, and did not read.

Conclusions: Our findings suggest that both poor lifestyle and occupational aluminum exposure may affect cognitive function. Workers must maintain a reasonable lifestyle and reduce aluminum exposure, which can control the occurrence of cognitive impairment.

Keywords: Bayesian networks; Cognitive impairment; Lifestyle; Occupational aluminum exposure.